Wavelet-based multi-resolution analysis can decompose a time series into a set of constitutive series with an explicitly defined hierarchical structure. This decomposition method can improve the accuracy of forecasts of original time series data. We apply it to expand the wheat yield for the states of Punjab, Haryana and Bihar in India during the 52-year period from 1966 to 2017 into a group of hierarchical series in a meaningful manner. The improvement in forecasting performance of the multi-step forecasts obtained using multi-resolution analysis is shown in terms of minimum values of mean absolute error and root mean square error. A comparative study for predictive performance is also conducted between the wavelet-based multi-resolution augmented method and the corresponding conventional approach, and the first approach is revealed to be better.
